打印
[少儿编程]

深度学习圣经:Deep learning

[复制链接]
139|2
手机看帖
扫描二维码
随时随地手机跟帖
沙发
keer_zu|  楼主 | 2024-5-17 10:05 | 只看该作者
If we had separate parameters  for each value of the time index, we could not generalize to sequence lengths not  seen during training, nor share statistical strength across different sequence lengths  and across different positions in time.

使用特权

评论回复
评论
keer_zu 2024-5-17 13:22 回复TA
The idea of parameter sharing manifests in the application of the same convolution kernel at each time step. 
发新帖 我要提问
您需要登录后才可以回帖 登录 | 注册

本版积分规则

1344

主题

12407

帖子

53

粉丝